SC Linux Workstations
- Introduction
- Access requirements
- Locations and availability
- Obtaining help
- Usage policy and guidelines
- Logging in
- Disk space management (Warning!)
- Locally installed software
- Launching remote software
Introduction
The TAMU Supercomputing Facility continues to provide its users access to several Linux workstations at two separate locations on campus. These workstations serve as physical terminals for remote (network) access to the supercomputers, as well as stand-alone platforms for interactive post-processing and visualization (graphical and GUI-based) tasks that would typically be slower to carry out on the heavily loaded, larger machines, especially over the network.
This page is intended to be a brief users' guide both for new as well as existing users of these Linux workstations. As a first step in troubleshooting problems, users should always consult relevant sections of this guide, as updated information will be posted whenever the need arises.
Access requirements
We require the use of AMS to activate logins on the workstations.
To obtain access:
- Apply for an account on any of the Supercomputing servers, if you have not already done so.
- Send a request to account@sc.tamu.edu, requesting access to the SCLAB workstations.
- When you have been notified that you are approved, activate the SCLAB resource in AMS. Read our instructions on using AMS for more details. Allow up to 30 minutes for your account to be accessible.
Locations and availability
Six Linux workstations (their hostnames are: perk2.tamu.edu, perk3.tamu.edu, ..., perk7.tamu.edu) have been installed in the Open Access Lab at the Wisenbaker Engineering Research Center (WERC) in room 024. Additional information about the lab may be found here.
One additional workstation (hostname: perk1.tamu.edu) has also been installed at the Student Computing Center (SCC). This lab is typically open 24 hours during the weekdays and for a more limited time on the weekends.
The daily schedule for both labs can be found here.
Obtaining help
The support staff at both labs mentioned above should be the initial point of contact for problems that are possibly hardware related. For most other problems, especially those relating to the OS and the installed software, users must contact the Supercomputing Helpdesk by calling 845-0219 or by sending email to help@sc.tamu.edu. When sending email, be sure to follow these recommendations so we can help you in a prompt and efficient manner.
Usage policy and guidelines
The Linux workstation accounts are not to be used for work that is unrelated to the purpose for which a user was granted a supercomputing account.
These machines are meant primarily for interactive tasks (post-processing or visualization work). Users who need to work primarily through the batch systems on the supercomputers (submitting batch jobs and running programs on the command line) may do so using any of the windows machines in the open access labs. They simply need a secure shell window to log in to one or more of the supercomputers.
When both types of users wish to use the Linux workstations, those with non-interactive needs must yield use of the machines to those who have interactive/graphical/visualisation tasks to perform.
Furthermore, no user may occupy a workstation for more than 2 hours in a single sitting, unless there is no one else waiting to use a workstation.
Logging in
Users may login to these workstations either at the console (while sitting in front of the workstation), or remotely over the network using a secure shell connection (SSH). In either case, the user must supply a username and password to gain access to the account.
You must activate your username/password for the Linux workstations via AMS. The Linux workstations are represented by the SCLAB AMS resource.
Disk space management
There are several directories on each workstation where users may store files. It is important to understand the characteristics of each directory. Storing your data in the "wrong" place could mean that you will need to expend time and effort to find it again, and in certain cases, it may mean losing your data altogether!
Please note that the /SCscratch directories listed in the
table below are 'automounted' on the workstations and consequently may not be
visible in directory listings produced by the 'ls' command or in the output of
commands such as 'df' until and unless they are actually used by the user (e.g.
a user cd's to one of these directories or copies a file to or from such a
directory).
| Directory Name | Space | Backed up? 1 | Persistent? 2 | Shared 3 Over Network? |
/tmphome/user |
45 MB | No | Yes | No |
/scratch/user |
16 GB max | No | No | No |
- Backed up data is that which is written to tape so that it
may be recovered in the event the copy on disk storage is
corrupted/destroyed. None of the directories listed above
are backed up to tape. It must be noted, however, that data
stored in the
/SCscratchfilesystem under thecosmossub-directory is RAID 5 storage. RAID 5 storage devices provide storage redundancy and are secure againt data loss due to single hard disk failures. This means that data stored in the RAID 5 directories, while it is not backed up to tape, is nevertheless quite secure. - Persistent storage means that data will not be deleted across
system re-boots, or by automated administrative scripts. In
some systems, certain directories such as
/tmpcan be "cleaned out" automatically by the OS when the system comes back up after a re-boot. Data in such "temporary" directories may also be deleted by administrative scripts which try to ensure sufficient empty space by periodically deleting older files in these shared disk areas. The/scratchfilesystem on each workstation is automatically cleaned up by an automated script whenever it fills up to 80% capacity. The script keeps deleting files (oldest first) until the filesystem is only 20% full. - Directories which are not shared, like
/scratch/aggieor/tmphome/aggiefor instance, represent separate and independent disk storage that is local to each workstation. Saving a file in/scratch/aggieon the workstation perk2 means that you cannot access it in/scratch/aggieon any workstation other than perk2. You might find that even when you later return to perk2 to access the file, it may have been deleted by the automated cleanup script (see note 2). You have been warned!
Locally installed software
The following software packages have been installed on each of the Linux workstations to enable users to offload postprocessing and visualization tasks from the heavily loaded (and therefore interactively less responsive) supercomputers.
In the course of maintaining these packages on the workstations, whenever we become aware of problems/issues with any package, we will post solutions/work-arounds to the problems in the section below.
| Software | Command |
|---|---|
| Abaqus 6.6 CAE | abq663 cae (1) |
| Abaqus 6.6 Viewer | abq663 viewer (1) |
| Abaqus 6.7 CAE | abq672 cae (1) |
| Abaqus 6.7 Viewer | abq672 viewer (1) |
| Abaqus 6.8 CAE | abq683 cae (1) |
| Abaqus 6.8 Viewer | abq683 viewer (1) |
| Ensight 8.2 | ensight8 |
| Gambit 6.3.16 | gambit |
| Matlab 2009b | matlab |
- NOTE: for users that are running Abaqus CAE and Viewer on a SCLAB workstation and using X11 forwarding to display the GUI on their Windows PC. The -mesa command line option should be used if your X11 emulator does not support OpenGL (eg. Xming).
Launching remote software
If there is a need to launch software that resides on one of the supercomputers but have it display it's graphical interface on the Linux workstation where the user is sitting, the following instructions can be used:
Assume the program name is "xemacs" and that it resides on and will be executed on the machine "hydra". At the prompt in a unix shell window (on your workstation), type:
ssh -X hydra
Enter your (hydra) password in response to the password prompt generated by the ssh program. Once you are logged in to hydra, type:
xemacs